A Novel Hybrid Deep Learning-based Approach for Sensor Data Recovery in Structural Health Monitoring

Abstract

Structural health monitoring (SHM) systems contribute significantly to ensuring the safety of construction works. However, in reality, data loss often occurs due to many different reasons. A unique hybrid deep learning-based method for recovering sensor data in structural health monitoring (SHM) is presented in this research. The suggested technique accurately reconstructs missing or corrupted sensor data by utilizing the advantages of both Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN). While the RNN models time dependencies to recover the missing sequences, the CNN pulls important patterns from the data. The method's great accuracy in recovering sensor data, even under complex circumstances, is proven using case study real-world bridge monitoring data. The steps taken and the analysis of the results are clearly stated in the study. According to the results, the CNN-RNN combination performs better than conventional techniques and provides notable reliability gains for SHM applications. Future studies will try to improve the model even further and investigate how it may be used to a variety of sensor data and structural types.